Design of a Sanskrit Devanagari Script Recognition and Sanskrit to Chinese Translation System Based on Deep Learning

This study endeavors to enhance the accuracy of Sanskrit Devanagari script recognition through advanced deep learning techniques and to achieve accurate conversion from Sanskrit to Chinese using machine translation technology, in support of the digitization of ancient literature and the preservation of cultural heritage. A combination of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) is used for text recognition, along with a neural machine translation system based on the Transformer model for translating Sanskrit to Chinese. Experimental results show that the OCR system employed in this study achieves an accuracy rate of 89%, a recall rate of 77%, and an F1 score of 0.826 in recognizing Sanskrit text. The machine translation system effectively converts Sanskrit into semantically accurate Chinese text, with translation quality quantified by BLEU and METEOR evaluation metrics, reaching scores of 41% and 30% respectively. In summary, the findings of this paper not only optimize the text processing workflow for Sanskrit but also provide a similar framework and practical experience for text recognition and translation of other similar languages, contributing to the further advancement of cultural exchange and linguistic research.